• file
• revisions
• annotate
• diff
• comparison
• raw

intl/icu/source/common/ruleiter.cpp@b8a032363ba2 (annotated)

intl/icu/source/common/ruleiter.cpp

Thu, 22 Jan 2015 13:21:57 +0100

author

Michael Schloh von Bennewitz <michael@schloh.com>

date

Thu, 22 Jan 2015 13:21:57 +0100

branch

TOR_BUG_9701

changeset 15

b8a032363ba2

permissions

-rw-r--r--

Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6

michael@0	1	/*
michael@0	2	**********************************************************************
michael@0	3	* Copyright (c) 2003-2011, International Business Machines
michael@0	4	* Corporation and others. All Rights Reserved.
michael@0	5	**********************************************************************
michael@0	6	* Author: Alan Liu
michael@0	7	* Created: September 24 2003
michael@0	8	* Since: ICU 2.8
michael@0	9	**********************************************************************
michael@0	10	*/
michael@0	11	#include "ruleiter.h"
michael@0	12	#include "unicode/parsepos.h"
michael@0	13	#include "unicode/symtable.h"
michael@0	14	#include "unicode/unistr.h"
michael@0	15	#include "unicode/utf16.h"
michael@0	16	#include "patternprops.h"
michael@0	17
michael@0	18	/* \U87654321 or \ud800\udc00 */
michael@0	19	#define MAX_U_NOTATION_LEN 12
michael@0	20
michael@0	21	U_NAMESPACE_BEGIN
michael@0	22
michael@0	23	RuleCharacterIterator::RuleCharacterIterator(const UnicodeString& theText, const SymbolTable* theSym,
michael@0	24	ParsePosition& thePos) :
michael@0	25	text(theText),
michael@0	26	pos(thePos),
michael@0	27	sym(theSym),
michael@0	28	buf(0),
michael@0	29	bufPos(0)
michael@0	30	{}
michael@0	31
michael@0	32	UBool RuleCharacterIterator::atEnd() const {
michael@0	33	return buf == 0 && pos.getIndex() == text.length();
michael@0	34	}
michael@0	35
michael@0	36	UChar32 RuleCharacterIterator::next(int32_t options, UBool& isEscaped, UErrorCode& ec) {
michael@0	37	if (U_FAILURE(ec)) return DONE;
michael@0	38
michael@0	39	UChar32 c = DONE;
michael@0	40	isEscaped = FALSE;
michael@0	41
michael@0	42	for (;;) {
michael@0	43	c = _current();
michael@0	44	_advance(U16_LENGTH(c));
michael@0	45
michael@0	46	if (c == SymbolTable::SYMBOL_REF && buf == 0 &&
michael@0	47	(options & PARSE_VARIABLES) != 0 && sym != 0) {
michael@0	48	UnicodeString name = sym->parseReference(text, pos, text.length());
michael@0	49	// If name is empty there was an isolated SYMBOL_REF;
michael@0	50	// return it. Caller must be prepared for this.
michael@0	51	if (name.length() == 0) {
michael@0	52	break;
michael@0	53	}
michael@0	54	bufPos = 0;
michael@0	55	buf = sym->lookup(name);
michael@0	56	if (buf == 0) {
michael@0	57	ec = U_UNDEFINED_VARIABLE;
michael@0	58	return DONE;
michael@0	59	}
michael@0	60	// Handle empty variable value
michael@0	61	if (buf->length() == 0) {
michael@0	62	buf = 0;
michael@0	63	}
michael@0	64	continue;
michael@0	65	}
michael@0	66
michael@0	67	if ((options & SKIP_WHITESPACE) != 0 && PatternProps::isWhiteSpace(c)) {
michael@0	68	continue;
michael@0	69	}
michael@0	70
michael@0	71	if (c == 0x5C /*'\\'*/ && (options & PARSE_ESCAPES) != 0) {
michael@0	72	UnicodeString tempEscape;
michael@0	73	int32_t offset = 0;
michael@0	74	c = lookahead(tempEscape, MAX_U_NOTATION_LEN).unescapeAt(offset);
michael@0	75	jumpahead(offset);
michael@0	76	isEscaped = TRUE;
michael@0	77	if (c < 0) {
michael@0	78	ec = U_MALFORMED_UNICODE_ESCAPE;
michael@0	79	return DONE;
michael@0	80	}
michael@0	81	}
michael@0	82
michael@0	83	break;
michael@0	84	}
michael@0	85
michael@0	86	return c;
michael@0	87	}
michael@0	88
michael@0	89	void RuleCharacterIterator::getPos(RuleCharacterIterator::Pos& p) const {
michael@0	90	p.buf = buf;
michael@0	91	p.pos = pos.getIndex();
michael@0	92	p.bufPos = bufPos;
michael@0	93	}
michael@0	94
michael@0	95	void RuleCharacterIterator::setPos(const RuleCharacterIterator::Pos& p) {
michael@0	96	buf = p.buf;
michael@0	97	pos.setIndex(p.pos);
michael@0	98	bufPos = p.bufPos;
michael@0	99	}
michael@0	100
michael@0	101	void RuleCharacterIterator::skipIgnored(int32_t options) {
michael@0	102	if ((options & SKIP_WHITESPACE) != 0) {
michael@0	103	for (;;) {
michael@0	104	UChar32 a = _current();
michael@0	105	if (!PatternProps::isWhiteSpace(a)) break;
michael@0	106	_advance(U16_LENGTH(a));
michael@0	107	}
michael@0	108	}
michael@0	109	}
michael@0	110
michael@0	111	UnicodeString& RuleCharacterIterator::lookahead(UnicodeString& result, int32_t maxLookAhead) const {
michael@0	112	if (maxLookAhead < 0) {
michael@0	113	maxLookAhead = 0x7FFFFFFF;
michael@0	114	}
michael@0	115	if (buf != 0) {
michael@0	116	buf->extract(bufPos, maxLookAhead, result);
michael@0	117	} else {
michael@0	118	text.extract(pos.getIndex(), maxLookAhead, result);
michael@0	119	}
michael@0	120	return result;
michael@0	121	}
michael@0	122
michael@0	123	void RuleCharacterIterator::jumpahead(int32_t count) {
michael@0	124	_advance(count);
michael@0	125	}
michael@0	126
michael@0	127	/*
michael@0	128	UnicodeString& RuleCharacterIterator::toString(UnicodeString& result) const {
michael@0	129	int32_t b = pos.getIndex();
michael@0	130	text.extract(0, b, result);
michael@0	131	return result.append((UChar) 0x7C).append(text, b, 0x7FFFFFFF); // Insert '|' at index
michael@0	132	}
michael@0	133	*/
michael@0	134
michael@0	135	UChar32 RuleCharacterIterator::_current() const {
michael@0	136	if (buf != 0) {
michael@0	137	return buf->char32At(bufPos);
michael@0	138	} else {
michael@0	139	int i = pos.getIndex();
michael@0	140	return (i < text.length()) ? text.char32At(i) : (UChar32)DONE;
michael@0	141	}
michael@0	142	}
michael@0	143
michael@0	144	void RuleCharacterIterator::_advance(int32_t count) {
michael@0	145	if (buf != 0) {
michael@0	146	bufPos += count;
michael@0	147	if (bufPos == buf->length()) {
michael@0	148	buf = 0;
michael@0	149	}
michael@0	150	} else {
michael@0	151	pos.setIndex(pos.getIndex() + count);
michael@0	152	if (pos.getIndex() > text.length()) {
michael@0	153	pos.setIndex(text.length());
michael@0	154	}
michael@0	155	}
michael@0	156	}
michael@0	157
michael@0	158	U_NAMESPACE_END
michael@0	159
michael@0	160	//eof